Dr.

Urooj A Siddiqui
 Data – Raw Facts, especially numerical facts,
collected together for reference or
information.
 Data is collected on some particular
variable/s
 Data analysis is processing of data to derive
useful information
 Knowledge communicated concerning some
particular fact
 The created knowledge helps in APPLICATION /
DECISION MAKING
 Categorical:Qualitative
 Continuous: Quantitative

Data

Categorical Continuous

Nominal Ordinal Interval Ratio

 Any phenomenon which takes at least two
different values/ observations

 Data:Set of values/ observations

collected on variable is called data
 Nominal
 Ordinal
 Interval
 Ratio
1. Data Preparation / Initial 2. Summarizing Data / Data
Operations Analysis Operations

 Tables / Crosstab
 Editing / Cleaning
 Graph / Figure
 Coding  Statistical Analysis
 Classification 1. Descriptive Methods
 Frequency, %age, Ratio,
 Tabulation
 Mean, Median, Standard
 Graphical Deviation (Variance)
Representation 2. Inferential Methods
 Comparison (t/z-test/Anova)
 Association (chi square test)
 Correlation (r)
 Prediction/ Regression
(y = ax + b)
 Editing / Data Cleaning
 examining the collected raw data to detect any errors
and omit/correct it if possible
 Coding
 assigning numerals to answers so that responses can
be put into a limited number of categories
 Classification
 Grouping of data on some basis (large volume of raw
data is reduced into homogenous groups
I. Attribute - on the basis of demographic bases
eg. gender, rural/urban, day scholar/hosteller
II. Class Interval – on the basis on some numeric range
eg. 0-10, 10-20 etc.
I. Tabulation
 is the process of displaying raw data in tabular
form and summarising it for further analysis
 orderly arranging data in columns and rows
Tabulation is essential because
 It conserves space and reduces statements
 It facilitates the process of summation of
items, comparison, detection of errors and
omissions
 Basis for various statistical computations
 temp of
Gende Yrs in Pain
Name Caste Age Mob. No. Edu IQ locality
r school level
deg cel

Ram M Hindu 60 9450366367 NIL 0 16 Mild-0 -4

Akbar M Muslim 65 8004896712 HS 16 14 Mod-1 20

Sita F Hindu 309 9934876545 Int. 19 0 Mild-0 15

Shalini F Hindu 90 2542543598 HS 8 16 Mild-0 0

Mehnaj F Sikh 38 9458098734 UG 21 13 Severe-2 0

Ravi M Hindu 48 9412890112 PG 23 20 Mod-1 -1

Hari M Hindu 45 8796654398 Prim 12 10 Mod-1 30

 temp of
Edu Yrs in Pain
Name Gender Caste Age Mob.No. IQ locality
level sch. level
deg cel

7 1 1 60 9450366367 1 0 16 0 4

2 1 2 65 8004896712 1 16 14 2 20

5 2 1 35 9934876545 2 19 0 0 15

4 2 1 90 2542543598 1 8 16 0 0

3 2 3 38 9458098734 3 21 13 3 0

6 1 1 48 9412890112 4 23 20 2 -1

1 1 1 45 8796654398 0 12 10 2 30

Nominal & Ordinal called qualitative . Interval and Ratio called quantitative
 Roll. Age
 Single / Multi Variable Table - one or No (yr)
more variable (no interaction) 1 22
2 24
Single Variable Freq. Table
3 23
Age Group (years) Freq.
4 26
Below 20 2
5 19
20-22 28
6 25
22-24 16
. .
24-26 10
. .
Above 26 4
. .
60 . .
. .
**Multiple Variable Table – as presented in above slide
60 22
 Crosstabs – interaction of two or more
variables
Two Variable Interaction – Crosstab

Gender

Age Group Male Female Total

Below 20 1 1 2
20-22 18 10 28
22-24 9 7 16
24-26 7 3 10
Above 26 3 1 4
38 22 60
Graphical Representation of Data
 Pie Chart
 Bar Graph
 Histogram
 Line Graph
 Scatter Plot
 Scatter Plot & Correlation
Pie Charts
 It is used to represent %ages, distribution of 1
variable at various levels

Sales (in mn)

1.2,
8%
1.4,
10% 1st Qtr
2nd Qtr
3.2, 8.2, 58% 3rd Qtr
23%
4th Qtr
Bar Chart
 It is used to represent 1 variable at various levels
 Levels can be year/ groups etc.

4 Sales
3.5
3
2.5
4.3 4.5
2
3.5
1.5
2.5
1
0.5
0
2018 2019 2020 2021
Bar Chart
5 Clustered Bar
4.5
4
3.5
3 1st
2.5 2nd
2 4.3 4.4
4 3rd
3.5
1.5 3 3 4th
1 2.4 2.5 2.5
2 2 1.8
0.5
0
2018 2019 2020
 Histogram
 To show the distribution of a Roll. Age
No (yr)
quantitative variable
1 22
2 24
3 23
12
4 26
10
5 19
8
Frequency

6 25
6
10 . .
4 8
6 . .
2 4 . .
2 0
0
10 20 30 40 50
. .
Class Interval/Variable Unit . .
60 22
 Line Diagram
 To show change in variable in a particular time
period / on some reference range

₹ 7.40

₹ 7.20

₹ 7.00

₹ 6.80
Stock Price

₹ 6.60

₹ 6.40

₹ 6.20

₹ 6.00

₹ 5.80

₹ 5.60
1 2 3 4 5 6 7 8 9 10

Last 10 Days
Line Diagram
 May also be used to compare 2 or more variables
along the range
14
12
10
8 Adani
6 Tata
4 Reliance

2
0
1 2 3 4 5 6 7 8
Scatter Plot
 It is used to express relationships between two
variables
6
5
4
Sales in
3
Crore
2 Y-Values

1
0
0 1 2 3 4
Adv Budget in 10’Lacs
Scatter Plot
 to express relationships between two variables
Scatter Plot
 Trend Lines - Correlation
 No. of
Income / day 80
families
70
0-500 20
60
500-1000 30
50

No.of families
1000-1500 50 40

1500-2000 70 30

2000-2500 40 20

2500-3000 30 10

3000-3500 10 0
0 1000 2000 3000 4000
Income
. .
 age (xi) x-xi (x-xi) sqr.
A 21 2 4
B 22 1 1
C 23 0 0
D 24 -1 1
E 25 -2 4
10 (sum x-xi sq)
mean x 23 Sum 0

Avg Sq (variance) 2 (10 by 5), n=5

SD (root v) s 1.41
Roll. Age
No (yr) Age Group (years) Freq. Probability
1 22 Below 20 2 2/60
2 24
20-22 28 28/60
3 23
22-24 16 16/60
4 26
24-26 10 10/60
5 19
Above 26 4 4/60
6 22
60
. .
Mean 23 (years)
. . (x-sample-known)
. . (µ-population - unknown)

. . SD 2 (years)
(s-sample-known)
. . (𝜎 – population - unknown)
60 22
A distribution in frequencies of observations is
known – probability distribution

 Z- Normal Distribution/Test - Mean (µ), SD-

 To compare means (1 or 2 means)
t – Distribution/Test- Mean (x), SD (s)
 To compare means (1 or 2 means)
 Chi Square Distribution / Test
 To compare sample SD with population SD
F Test
 To compare two sample variances
A freq. distribution with bell shape curve and
some known properties
 Parameters - Mean (µ), SD (sigma)
 Known properties
 68% values are within µ ± 1 SD
 95% values are within µ ± 2 SD
 99% values are within µ ± 3 SD

 95% CI = µ ± 2.SD (range)

 Lower limit µ - 2.SD
 Upper limit µ + 2.SD
 23

21 25

19 27

17 29
Example of our case
 95% CI = µ ± 2.SD
 Lower limit = µ - 2.SD, Upper limit = µ + 2.SD,
 LL = 23 - 2.2 = 19, UL = 23 + 2.2 = 27
 95% CI Range = 19-27 years
 95% of the students in the class are in the range
of 19-27 yrs
 We are 95% confident that if we randomly select
a student from the class his/her age will be
within this range (19-27 yrs)
 Reverse is Hypothesis Testing
 If mean and SD of any population is known and if
some value is given can we determine whether it
belongs to this population or distribution ?
 0

-0.5 +0.5

-1
+1

-1.5 +1.5
Finding Probability
 Calculate z score (test statistic) of the observed
value or hypothesized value with the formula
 Determine p value associated with particular z
score at selected significance level (5%)
 P value can be seen in the tables of the particular
test

When Population SD is KNOWN When Population SD is UNKNOWN

t=
 Two types of Hypothesis, Null - H0, Alternate - Ha
P Value Method Table Value Method
 Determine p value  Calculate test statistic

 Compare with selected value – TSCal

alpha level (0.05)  Determine Critical value

 p ≤ 0.05 – Reject Null of test statistic at

selected significance level
 P > 0.05 – Fail to Reject
– TSTab
null / accept null
 If TSCal ≥ TSTab – Reject
 This method is generally
Null
employed by data analysis
software – Excel, SPSS  If TSCal < TSTab – Fail to
Reject null / accept null
 This method is generally
employed when manual
testing is done
 No. of Marks Specialization
Gender Caste Age
RN Mob.No. Classes Obtained Opted
G C A
N M S

1 1 1 22 9450366367 87 72 HR-3

2 1 2 24 8004896712 65 68 HR-3

3 2 1 26 9934876545 48 56 Fin.-2

4 2 1 21 2542543598 95 83 Mktg.-1

5 2 3 22 9458098734 65 58 Fin.-2

6 1 1 23 9412890112 74 65 Mktg.-1

• Mean & Variance (SD) – Eg. A, N, M – sample stat. – x, s

• Correlation Eg. N-M, A-N, A-M –r
• Association between Gender and Sp. Opted (G n S) - chi
Note Sample Ch.c – Statistic , Population Ch.c - Parameter
 Assume a population – N, µ,
 Now assume we take many samples of size n and
calculate mean for each sample
 x1, x2, x3, x4, x5, x6, . . . . . . . . x100
 Can we make a freq. distribution of these values
and draw a curve?
 Now when we draw a distribution of these values
we will have an average (x) and SD (s)
 This average is called mean of means and
considered mean of population
 The SD of population is calculated as
which is called as Standard Error
 Sample mean & their difference - z / t
 Sample correlation statistic– z / t (derived from r)
 Variance (SD2) – F
 Association – Chi Sqr.

 Central Limit Theorem

 If we collect many samples and draw its
distribution the mean of this distribution is
population mean and SD of population is
 We use CLT in Hypothesis Testing
z - when is Known and sample size is ≥ 30
 t - when is Unknown and sample size < 30
 In sample estimation t test is employed

 Example - H0 & H1
 H0 – There is no difference b/w mean of two groups
 H1 – There is a significant difference b/w mean of two groups
 H0 – There is no difference b/w mean marks of males &
females
 H1 – There is a significant difference b/w male & females
 Hypothesis Testing steps
 Set Null Value (u1=u2, u1-u2=0) – Make Null Distribution –
Calculate z /t sample test statistic – compare with table
value/set p value – reject/accept null
 Used to compare variance of two samples
 Employed in ANOVA – analysis of variance
 When there are more than two groups and their
means are to be compared
 Example
 Comparison of marks among three streams of
students arts, commerce and science
 H0 – There is no difference among mean marks of three groups
 H1 – There is a significant difference among mean marks of three
groups

 Set Null Value (µ1=µ2=µ3) – Make Null Distribution – Calculate F

test statistic – compare with table value/p value – reject/accept
null
 Test of Independence

 It is used to determine association between two
categorical variables (nominal & ordinal)
 Example
 Gender (M/F) and Opted Specialization (M/F/HR)
 Question like ‘is any specialisation is preferred by
females?’ are answered
 H0 – There is no association b/w gender and opted speclisa.n
 H1 – There is a significant association b/w gender & opted
speclisa.n
 Here, mean is not calculated instead frequency of categories
is taken into consideration
 Actual Frequency and Expected Frequency
 Cross tabs are used to calculate actual & expected freq

Two Variable Interaction – Crosstab

Opted Total Gender

Specialization (60) Male (40) Female (20)
Mktg. 30 20 8
Fin. 15 10 2
HR 15 10 10
60 40 20

 Hypothesis Testing steps

 Set Null Value (actual freq. = expected freq.) – Make Null
Distribution – Calculate chi sqr. sample test statistic –
compare with table value/set p value – reject/accept null
 Set Null and Alternate Hypothesis – H0 H1
 Select the null value
 Null – status quo, no difference, no effect
 Status quo – no change
 No difference – 0 difference
 No relationship – 0 effect / 0 correlation
 No association – 0 relationship (b/w nominal variab.)
 It is assumed that H0 is true in population
 Draw Null Distribution – find range of expected values
if null is true (µ ± 2.SE)
 Take observed value from sample and compare with
expected null values
 If observed value is among expected null range –
accept null
 If observed value is different from null range – reject
null
1. Univariate/Bi-variate 2. Muti-variate

 Mean/Variance  Correlation
Estimation  Regression
 Z test  Discriminant
 T test  Cluster Analysis etc.
 Chi Square
 F Test
 Correlation
 Regression analysis
 1 dependent variable/DV (continuous)
 many independent variables/IV (continuous)
 Y = a.x1 +b.x2 +c.x3…….+.x.n

 Discriminant analysis
 1 dependent variable (categorical)
 many independent variables (continuous)
 Z (yes/no) = a.x1 +b.x2 +c.x3…….+.x.n
 Cluster analysis
 No DV/IV
 Used to group respondents/customers in
various cluster
 Employed in market segmentation

 Factor analysis
 No DV/IV
 Used to group variables in various cluster of
more condensed variables